On 14/10/11 11:01, Earl Ruby wrote:
I have been using the monolithic version of Heartbeat for a few years
and I'm working on setting up a Corosync/Pacemaker cluster. I'm
currently using two OpenSUSE 11.4 virtual machines under Xen to test
with and I'm using the standard OpenSUSE packages.
On a freshly-booted server there are <defunct> processes under corosync:
10859 ? Ssl 0:00 /usr/sbin/corosync
10865 ? S 0:00 \_ /usr/lib64/heartbeat/stonithd
10866 ? S 0:00 \_ /usr/lib64/heartbeat/cib
10867 ? Z 0:00 \_ [lrmd] <defunct>
10868 ? S 0:00 \_ /usr/lib64/heartbeat/attrd
10869 ? Z 0:00 \_ [pengine] <defunct>
10870 ? Z 0:00 \_ [crmd] <defunct>
10871 ? Z 0:00 \_ [mgmtd] <defunct>
10873 ? S 0:00 \_ /usr/lib64/heartbeat/stonithd
10874 ? S 0:00 \_ /usr/lib64/heartbeat/cib
10875 ? S 0:00 \_ /usr/lib64/heartbeat/lrmd
10876 ? S 0:00 \_ /usr/lib64/heartbeat/attrd
10877 ? S 0:00 \_ /usr/lib64/heartbeat/pengine
10899 ? S 0:00 \_ /usr/lib64/heartbeat/crmd
Note that although there are defunct lrmd, crmd, pengine and mgmtd
processes there are also there are also running copies of all of these
except mgmtd.
This problem has been reported to Novell/SUSE:
https://bugzilla.novell.com/process_bug.cgi, but no solutions were offered.
What's the bug number please?
Can anyone tell me why this is happening or suggest configuration
settings I should look at to keep this from happening?
This looks suspiciously like the symptoms of having libheartbeat2
installed instead of libglue2 (although that shouldn't be possible on
openSUSE 11.4, I thought). See:
https://bugzilla.novell.com/show_bug.cgi?id=628243
In any case, it's difficult to say for sure without seeing
/var/log/messages around the time corosync was starting.
Regards,
Tim
--
Tim Serong
Senior Clustering Engineer
SUSE
tserong@xxxxxxxx
_______________________________________________
discuss mailing list
discuss@xxxxxxxxxxxx
http://lists.corosync.org/mailman/listinfo/discuss
